Agentic commerce is a model of digital shopping in which AI agents act on behalf of consumers to discover products, evaluate options, and complete purchases — without requiring the buyer to navigate a website, fill out forms, or manually initiate checkout. The transaction happens inside the AI interaction itself.
The human sets intent, approves decisions, and retains control. The agent handles execution. A shopper expresses what they want in natural language, the agent interprets that intent, acts across merchant systems, and presents options or completes a purchase depending on permission level and transaction type.
This commerce-specific autonomy is distinct from broader agentic AI capabilities. Agentic AI is the general ability of systems to reason, plan, and act. Agentic commerce is the specific application of that ability to buying and selling — an agent with purchasing intent, a structured way to discover and interact with merchants, and a payment and trust layer both sides can rely on. Autonomy varies, from price alerts and comparisons to end-to-end purchasing.
The Core Idea: From Discovery to Purchase Inside a Single Conversation
| Shopping Phase | Legacy Web Experience | Agentic UCP Flow | Underlying AI Surface | Trust & Verification | Conversion Velocity |
|---|---|---|---|---|---|
| Discovery & Research | Browsing 10+ tabs & blog reviews | Synthesized multi-source comparison in-dialog | AI Mode in Google Search | Verified Google Shopping Graph | 10x Faster Decisions |
| Variant Customization | Dropdown clicking & size charts | Conversational voice & vision refinement | Gemini Live API (Multimodal) | Real-time stock webhook | Zero Sizing Fatigue |
| Cross-Brand Cart | Multiple logins & independent checkouts | Unified cross-merchant Universal Cart | UCP Universal Cart Engine | Merchant of Record isolation | Single Action Buy |
| Payment Settlement | Entering credit cards on merchant sites | One-click biometric authorization inside chat | Google Pay (PAN_ONLY) | Tokenized zero-PII settlement | +42% Conversion Lift |
| Post-Purchase Care | Support tickets & telephone queues | Direct conversational returns & live tracking | Autonomous Customer Agents | Signed transaction hash | Instant Resolution |
| Replenishment | Forgotten passwords & lost carts | Contextual replenishment prompts | Background Reasoning Swarms | Recurring spending tokens | +65% LTV Retention |
In traditional e-commerce, a shopper searches for a product, clicks through to a website, browses, selects a size or variant, adds to cart, enters shipping information, enters payment details, and confirms the order. Each step adds friction.
In agentic commerce, many of those steps collapse into the conversation itself. A shopper interacting with Google’s AI Mode in Search or the Gemini app can express need in natural language. The AI surface discovers relevant products, presents options, and — for merchants who have integrated with the Universal Commerce Protocol (UCP) — can initiate and complete a checkout session without the shopper leaving the conversation.
For example: “find me running shoes under $150 with good arch support that I can get delivered by Friday.” The agent handles the workflow of querying merchant catalogs, filtering by price and specifications, and checking inventory and delivery estimates, then presents the best options. If the shopper says “buy the second one,” the agent completes the transaction using saved payment credentials — no checkout form, no account creation in the moment.
Or: “Find me a camping tent under $150 and have it delivered by Friday.” The human declares intent. The AI agent interprets and acts. It accesses structured product data, filters by price, specs, and delivery availability, compares options across merchants, and then presents recommendations or completes the purchase.
The shopping cart was invented because browsers needed a mechanism to accumulate items before a single payment transaction. It is a technical artifact of the HTTP request-response model, not a natural expression of shopping intent. In a conversational interface, the agent maintains the full context of what the shopper wants and can execute multi-item purchases in a single transaction.
In this model, payment credentials stored in Google Wallet and shipping addresses already on file can be applied within the agent-mediated transaction. The shopper confirms; the order is placed. The merchant remains the Merchant of Record.
What Makes an AI Agent a Commerce Agent?
An AI agent, in the context of commerce, is software that can take actions — not just generate responses. A commerce agent specifically can:
- Discover a merchant’s available products and capabilities by fetching the merchant’s UCP profile
- Negotiate which checkout capabilities, payment handlers, and fulfillment options are mutually supported
- Create and manage a checkout session on behalf of the buyer
- Apply discounts, loyalty credentials, or subscription terms within the checkout flow
- Complete the transaction programmatically when all required information is present
- Hand off to the buyer when human input is required — and resume when the buyer has provided it
The key distinction is that an agent acts. It does not simply return a list of links. UCP is the protocol that defines how agents and merchants communicate so those actions are possible.
How a Purchase Executes
At its core, agentic commerce requires three things: an agent with intent, structured discovery, and a payment and trust layer. The agent does not browse a website the way a human does. It calls APIs. It reads structured data. It operates through machine-to-machine protocols, not visual interfaces.
The process follows a structured sequence:
1. Discovery
The agent queries a merchant’s /.well-known/ucp endpoint to understand what the store offers and how to interact with it. This is the machine-readable equivalent of a shopper visiting a store’s homepage — a standard place to learn capabilities, checkout paths, and supported flows.
2. Capability Negotiation
The agent and the merchant’s system exchange information about what each side supports. The merchant declares its checkout paths, payment handlers advertised via ucp.payment_handlers, and fulfillment options. The agent declares what credentials and payment methods it can provide on behalf of the shopper. Only mutually supported combinations proceed.
3. Product Query
The agent searches the merchant’s catalog to see current pricing, inventory, and product details, then compares options. This keeps the conversation grounded in what the merchant can actually fulfill.
4. Checkout Session
When the shopper signals purchase intent, the agent initiates a checkout session — a structured transaction that calculates totals including tax and shipping, processes payment through the selected handler, and confirms the order. Implementations reference dev.ucp.shopping.cart and dev.ucp.shopping.checkout capability naming. Capability names and behavior are versioned; check the spec version your merchant and agent support.
5. Post-Purchase
The agent can support order status tracking, returns initiation, and ongoing order management through the same protocol, preserving session context across the lifecycle.
The Role of the Universal Commerce Protocol in Agentic Commerce
Agentic commerce requires a shared language. Without one, every AI platform would need a custom integration with every merchant, and vice versa — an impossible scaling problem. UCP is that shared language. A merchant that implements UCP once is accessible to every AI agent that supports the protocol. An AI platform that supports UCP can interact with every UCP-compatible merchant.
As Ashish Gupta, VP/GM of Merchant Shopping at Google, stated at launch: “The shift to agentic commerce will require a shared language across the ecosystem — and the Universal Commerce Protocol provides that framework. Through our collaboration with partners like Shopify we’re ensuring the protocol is interoperable and ready to meet the evolving demands of retailers and their customers.”
UCP handles discovery of what a merchant supports, negotiation of which capabilities and payment handlers both the agent and merchant share, the structure of the checkout session, and post-purchase order tracking — all through a single standardized protocol. Full technical documentation is available at ucp.dev.
On transports and versions: REST is the recommended default, with MCP available as Early Access Beta requiring onboarding. The current spec is v2026-08-25, while Google surfaces teach 2026-04-08 — expect a version lag between spec and surfaces and confirm supported versions during integration.
Where Agentic Commerce Is Live Today
Google AI Mode in Search
Google is enabling checkout on eligible product listings in AI Mode in Search. Shoppers in the United States can check out from eligible retailers directly within the AI Mode experience, using payment methods and shipping information already saved in Google Wallet. Retailers remain the Merchant of Record.
UCP-powered checkout became available for direct checkout for select U.S. retailers including Etsy and Wayfair, with first checkouts from February 2026. On October 15, 2026, Universal Cart went live across Search and the Gemini app with Target, Walmart, Wayfair and Shopify merchants participating.
Gemini App
The same UCP-powered checkout capability is being rolled out in the Gemini app, including the Gemini web app. Universal Cart coverage extended across Search and the Gemini app as of October 15, 2026, with single-item support in Canada from November 4, 2026.
Microsoft Copilot
Shopify’s Microsoft Copilot integration includes an embedded checkout experience called Copilot Checkout. This allows shoppers to purchase directly inside Copilot without leaving the conversation.
ChatGPT
OpenAI introduced native checkout in ChatGPT in September 2025 via its Agentic Commerce Protocol co-created with OpenAI, built with Stripe, live for U.S. Etsy at launch with Shopify following. In early 2026, OpenAI shifted toward merchant-controlled, discovery-led patterns — acknowledging merchants need to retain control over customer relationships and brand experience. Standalone Instant Checkout was deprioritised on March 24, 2026 to focus on discovery. Shopify manages its ChatGPT commerce integration through Shopify’s Agentic Storefronts, the centralized admin hub for agentic commerce channel management, on by default for eligible stores.
What Happens When a Transaction Requires Human Input?
Not every transaction can be completed autonomously. A furniture retailer may require a buyer to select a specific delivery date and time window. A merchant may require confirmation of final sale terms. In these cases, UCP’s design does not fail — it routes gracefully.
When human input is required, the merchant response provides structured context and a continuation URL. The buyer follows that URL and picks up where the agent left off — with session state preserved.
As Vanessa Lee, VP at Shopify, noted: “In the instances where a customer’s interaction is required — take a furniture retailer that requires a customer to select a specific delivery date and time — UCP gives a standard way for that furniture retailer to specify to agents what information is needed from customers in the checkout flow.”
Who Owns the Customer Relationship in Agentic Commerce?
This is one of the most important questions merchants ask. In UCP-powered agentic commerce, the answer is clear and confirmed in official documentation: the merchant remains the Merchant of Record.
Google’s official developer documentation states explicitly: “You remain the Merchant of Record. Keep all of your customer data and relationships.” The merchant retains full ownership of the customer relationship, transaction data, and post-purchase experience. The AI surface facilitates the transaction; it does not own it.
What Agentic Commerce Means for Merchants Practically
For consumers, shopping becomes faster and less frustrating. Agents can compare prices across merchants, remember preferences, and handle the mechanical work of completing a transaction.
For merchants evaluating whether and how to engage, the practical implications are:
- New reach without rebuilding checkout. Merchants can transact inside Google AI Mode, Gemini, and other AI surfaces without replacing existing checkout infrastructure. UCP integrates with existing business logic. Merchants on Shopify with native UCP support are accessible to UCP-compatible agents where eligible and enabled; others implement directly.
- Support for complex commerce. UCP handles discount codes, loyalty credentials, subscription billing cadences, and final-sale or pre-order confirmation within the agent-mediated checkout flow.
- Payment flexibility via handlers. Merchants advertise which payment handlers they accept via ucp.payment_handlers; agents select a compatible one dynamically per transaction. The payment layer is modular and open, so providers can participate via handlers without being baked into the core protocol.
- Discovery depends on structure. If product data is not structured for machine consumption, agents will struggle to find and compare it. Merchants preparing now focus on structured product data in UCP-compatible schemas, agent-accessible checkout flows, and machine-readable policies for shipping, returns, and warranties. Merchants who wait risk being invisible to agents that rely on structured APIs.
- Early access via waitlist and eligibility. Integration with Google’s AI surfaces requires eligibility and approval. Eligible stores may be on by default; others can request access via Google’s UCP interest form.
Merchant Voices on Agentic Commerce
Several merchants have spoken publicly about participation. These attestations are shared as testimony:
Victor Tam, CEO and Co-Founder of Monos: “At Monos, we’re excited about agentic shopping because it enables us to meet customers where they already are. It’s a new way for our story and product details to show up at the exact moment someone is asking real questions with real intent, in a format that feels helpful, not intrusive.”
Sam Buckingham, Director of Global Digital Product at KEEN Footwear: “We’re excited to partner with Shopify as innovators in AI-driven commerce, which we believe represents the future of how people shop.”
Consumer Agentic Shopping FAQ
In what ways is agentic shopping fundamentally different from chatbot recommendations?
Chatbot recommendations provide static text links that dump consumers onto landing pages. Agentic commerce executes complete transactions—checking live inventory, bundling items, and running payment—without leaving the conversation.
How does the Universal Cart facilitate multi-merchant purchasing in a single session?
The Universal Cart aggregates items across disparate retail domains into a unified session, routing payment and fulfillment sub-orders to each respective merchant under the UCP protocol.
Which platform surfaces currently support live agentic checkout?
Google’s AI Mode in Search and Gemini (web and mobile) have live direct checkout in the US (launched October 15, 2026) and Canada (launched November 4, 2026 for leading retailers).

Leave a Reply